I’ve been following baseball and baseball analytics fairly deeply for about six or seven years now. I like to think I could dig up stats, evaluate players, and talk about trends around the league. Six to seven years doesn’t seem like that long, or at least, not that long of a time frame for things to change. However, as you’ll notice in the lack of brevity in this article, things have changed a lot on the data front in that span.
So in case you just came out from under that proverbial rock, planning on renewing your MLB TV package for the first time in years, or just want a quick refresher, I’ll try to be your guide. We’ll touch on the basics, dive a little deeper into more nuanced ideas, talk about the new technology and trends if there is time (spoiler: there will be). I’ll also add a little bit of personal color to my thoughts on the stat and the best way to use them (ignore me at your own pleasure).
This won’t be all-encompassing, that would be impossible, or cumbersome, or both. But I will try to make it detailed, covering some of the things you’d see in baseball analytics along the way. There will be links as well to go deeper into subjects if you want to dive deeper down that rabbit hole.
Earned Run Average (ERA) - Everyone is likely familiar with this one and it’s the landline of pitcher stats: old, outdated, too simplistic. Your parents probably use it, and most people use something else. ERA was the gold standard at one point but it is now just one small part of pitcher evaluation. It’s a simple expression: the number of runs a pitcher would allow, based on how many they have, if they pitched for nine full innings. I could spill two-tons of digital ink on why ERA is bad but I don’t get paid by the word.
Earned Run Average Minus (ERA-) - While I would argue this is still not a great metric to use, it’s better than just using raw ERA. You’ll see this as a common method of expression for advanced stats, a + or - symbol. This simply is just ERA indexed to that year, with 100 being average and anything below (above) being better (worse). This makes ERA useful to compare across seasons, decades, and eras like the dead-ball era or the new rabbit ball years.
Fielding Independent Pitching (FIP) - The new standard of pitcher evaluation, though the “science” behind it is actually twenty years old. Back in 2001 (!!!), Voros McCracken first wrote about what pitchers might have control of. This would lay the groundwork for Defense Independent Pitching Statistics (DIPS), or the idea that when the ball leaves a pitcher's hand, they no longer have control of where it might end up. Be it a single, a double, or a triple, that is out of the control of the pitcher and is dependent in part on the skill and positioning of his teammates. FIP is the way to turn DIPS into a metric similar to ERA. It removes anything besides walks, strikeouts, home runs (sometimes hit by pitches are counted). Those events do not require seven fielders behind the pitcher to change the outcome. In any given year, the league average for FIP and ERA are the same, so they are comparable in-season.
Fielding Independent Pitching Minus (FIP-) - Just like ERA-, FIP- is just the year-indexed version of FIP. Lower is better.
Exit Velocity - A measurement of how fast the ball comes off the bat. Exit velocity can be used for both hitters and pitchers and obviously higher (lower) is better (worse) for a batter (pitcher). Anything over (under) ~91 MPH is considered good (bad) for a hitter (pitcher).
Soft/Medium/Hard Hit% - Using exit velocity, we can define a ball in a category of soft, medium, or hard. Hitting the ball hard is good for hitters, as hard-hit balls often are hits over the long run. Obviously, the opposite is true for pitchers.
Launch Angle - An incredibly simple idea that wasn’t really ever measured or discussed until only in the past couple of years (okay, true, Ted Williams was talking about it 60 years ago). The attack angle of the batter's bat will often determine the pitch of the batted ball. Too high of a launch angle will lead to pop flies that won’t leave the infield. Too low of a launch angle will lead to groundballs that won’t either. There is no universally perfect launch angle because different hitters should have different approaches. Hitters with a lot of raw power should try to lift the ball (or as it’s called now “elevate and celebrate”). Hitters who are more the speedy/low-power types should have a lower angle to try to beat out grounders and find holes as fly balls would often just lead to outs.
Barrels - So now that we have an exit velocity and launch angle, we can define a barreled ball. A ball that is barreled has some combination of exit velocity and launch angle that historically has had an outcome of a batting average >=.500 and a slugging percentage >=1.500. These types of batted balls often lead to not only hits, but extra bases and often home runs.
Sweet Spot - Not quite as useful or important as a barreled ball, but this refers to any ball hit at between an 8 to 32 degree launch angle, or what is considered the “sweet spot” on the bat.
Strikeout% (K%) - Simple one but very useful for determining both who is good in the majors and minor leagues. Minor league hitters with high strikeout rates often struggle to find success in the majors while pitchers with low strikeout rates struggle in the majors. That’s not to say a low (high) strikeout rate is automatically good for a hitter (pitcher) but the danger zone on the other end is important. Also when comparing to pitchers (ie: a starter vs reliever), using K% is a better metric since it isn’t dependent on how many innings a pitcher has pitched like strikeouts per 9 (K/9).
Walk% (BB%) - The reverse is basically true for everything I said about K%. High BB% for pitchers is bad mostly. High BB% for hitters is good (this one is universally good). It’s another good metric to figure out how a minor league may translate to the majors.
Whiff% - If a batter swings the bat and misses, that is a whiff. Whiff% is simply the percentage of times the batter swung the bat and missed. This is a nice indicator for pitcher success.
Swinging Strike% - This may sound similar to whiff% but it’s slightly different. Instead of the number of swings in the denominator, it’s the number of strikes. So this is the number of times a batter swings and misses as a percentage of strikes. Or to think about it another way: whiffs / total strikes. Again, a great metric in my opinion for finding minor league to major league success. This number has trended upwards as strikeouts have gone up.
Isolated Power (ISO) - This is actually a fairly old stat (Branch Rickey used it!) but it’s part of the new-age of analytics. ISO is the number of extra-base hits per at bats ultimately. It tries to go beyond slugging percentage, since slugging percentage (which includes singles) and batting average are mostly linked. ISO is simply slugging percentage minus batting average. Two batters can have the same slugging percentage but have very different batted ball outcomes.
Four singles and no home runs in ten at bats = .400 batting average and ,400 slugging percentage but a 0.000 ISO
Zero singles and one home run in ten at bats = .100 batting average and .400 slugging percentage but a .300 ISO
If the idea is to see who hit for more power (aka no hitting singles), isolated power is better than slugging percentage.
The Plate Disciplines and Batted Balls
A note on these: there isn’t really a perfect number for what you want from any hitter. It will depend on who the hitter is. I’ll give a few examples as we go along.
Pull% - The percentage of batted balls that were pulled. For power hitters, this is a good thing but it also means a batter may be more reliant on pitches inside the zone and struggle on outer half stuff. The average pull% is ~40%.
Center% - The percentage of batted balls hit to the center part of the field. No real good or bad here in my opinion, other than hitters who can hit towards the middle of the field are often those that have a bit better bat control and aren’t reliant on feasting on inside stuff necessarily. The average center% is ~35%.
Opposite% - Again, this isn’t one necessarily that is good or bad, it is just a quantifiable way to demonstrate what type of hitter someone may be. Opposite field hitters are typically lower power hitters, so if you have a hitter who demonstrates opposite field power, that’s a really good sign. The average opposite% is ~25%.
Soft% - The percentage of batted balls that were designated as softly hit. The velocity that determines what “soft” mean varies by who is providing the data, sometimes just being a complete judgment by whichever stringer is entering the data. For hitters, you obviously want a lower soft% as batted balls that turn into outs correlates with soft hit balls. As a pitcher, the more the better. League average is ~17%
Medium% - You guessed it; the percentage of batted balls are designated as medium hit. League average is ~53%.
Hard% - Like, soft% this can actually tell us something. A hitter (pitcher) with a high (low) hard% is typically a good (bad) thing. However, not all batted balls are alike obviously. These quality of contact metrics completely ignore the launch angle of the ball, so a hard-hit ground ball isn’t as good as a hard-hit line drive or fly ball. League average ~30%.
O-Swing% - The percentage of swings a batter takes on pitches outside the zone. The denominator here is outside pitches, not swings. It’s not categorically true that an ideal O-Swing% is 0% (aka never swinging at a pitch off the plate) but it’s probably mostly right. League average is ~28%
Z-Swing% - The percentage of swings a batter takes on pitches in the zone (denoted by Z here). This one is tough to narrow down exactly what is good or bad, as it depends on the other traits of the hitter. Some players need to swing more in the zone (something that Yoan Moncada did and took a giant leap forward) because they give away free strikes that lead to deep counts (and strikeouts). Obviously the more you swing overall, the less you walk, so there is some give and take here. Not to mention all pitches in the zone aren’t the same, and some selectivity here is good, as there are good and bad pitches in the zone. Average is ~65%.
Swing% - The percentage of pitches a batter swings at. There is some signal in the noise for guys who swing a lot also strike out a lot, but most importantly it matters what happens the moment after they swing (ie: if they make contact or not). Average is ~45%.
O-Contact% - When a batter swings at a pitch outside the zone, how often do they make contact? For pitchers you want this lower, as it means batters are swinging at these pitches and not making contact. For hitters, you mostly want this high but you want the denominator (swings at outside pitches) to be low, if that makes sense. ~66% is average.
Z-Contact% - Much like O-Contact%, this is just with pitches inside the zone. It works basically the same as O-Contact% in what you want from a hitter (you obviously never want the batter to make contact when they swing). For hitters, it’s complicated as well, but a high Z-Contact% is mostly good. ~87% is average.
Contact% - Take basically all the above and roll it into one and you’ve got contact%, or how often a batter makes contact when they swing. High contact rates are mostly good, but who the batter is matters a lot. A low-power hitter you want high contact rates with because they are a high-volume type of hitter where you want the batted balls to fall your way. For power hitters, you can live with lower contacts if sometimes when they do swing, the ball goes far. ~80% in average.
Zone% - I mostly use this as a fun/interesting stat as it can tell you one thing pretty clearly: do opposing pitchers fear the hitter? Or does the pitcher fear the opposing hitter. If a hitter has a high zone%, it often means the pitcher is okay with just pounding the zone because they don’t really fear the hitter. Now, some pitchers are just natural strike throwers (Max Scherzer, Cliff Lee, Phil Hughes) and others live on the fringe of the zone with great command (Zack Greinke)...and some pitchers can’t find the zone at all. Average is ~45%.
The Little More Advanced
Let’s talk about linear weights really quickly before diving into the next ones, as that is fundamental to understanding the below metrics. You might know On Base + Slugging (OPS) which is just obviously those two metrics added together. What that assumes though, is that every point of OBP is worth the same as every point of slugging. In reality, every point of OBP is worth ~1.8x as many points as slugging. What this means is that OPS tends to underrate (overrate) high on base (power) hitters. What linear weights do is find the true value of every type of event - from a walk to a hit to a home run to a strikeout - and weights them “correctly.”
Weighted On Base Average (wOBA) - Using linear weights as described above, this turns the traditional OBP into the same idea of a metric but with those linear weights described. It’s one of the fundamental concepts/stats in sabermetrics. wOBA changes every year essentially, to count for the environment of offense.
Weighted Runs Created Plus (wRC+) - There is actually a step between wOBA and wRC+ (called weighted runs above average - or WRAA) but we can skip past that for these purposes. Think of wRC+ as simply wOBA (ie: linear weights) but with league and park factors applied and then everything scaled to 100 being average. Every point above 100 is equal to 1% better than average for a hitter. Every point below, 1% below average. So a hitter with a 120 wRC+ is 20% better than average, a hitter with an 80 wRC+ is 20% below. Remember that this is adjusted by year/era, so we can compare hitters between years and decades, be it in the dead, live, or rabbit ball era.
Deserved Runs Created Plus (DRC+) - For several years wRC+ (a metric found exclusively on FanGraphs) mostly reigned supreme. Recently Baseball Prospectus rolled out a different way to measure hitters using a similar scale and idea, DRC+. This metric is a bit more granular than wOBA/wRC+, as where those two just look at the outcome of the play, DRC+ tries to look deeper, taking into consideration things like who the opposing pitcher is. Whereas older models used a “With or Without You” model (WOWY), DRC+ (and DRA- below) uses a mixed model approach, which accounts and neutralizes various variables all at once. A home run isn’t just 100% on the pitcher or hitter (like a WOWY or traditional approach may make it), just like a strikeout isn’t fully on one or the other either. Like wRC+, 100 is league average and points above/below represent percentages better/worse than that.
Deserved Runs Averaged Minus (DRA-) - This is essentially the pitching version of DRC+, attempting to control for many things in the individual event (hitter, pitcher, count, ball park...and even the umpire calling the game) rather than just the outcome. Where it does differ is that it is the opposite scaling wise as WRC+/DRC+. A number below average (100) is good (ie: 80 DRA- is 20% better than league average).
Ultimate Zone Rating (UZR) - Honestly, there is a lot to describe here, so I suggest just going to read FanGraphs primer on the subject if you want the heavy details. I’m just going to borrow the intro to that article and if you are curious for more, click the link I cited.
...UZR is an advanced defensive metric that uses play-by-play data recorded by Baseball Info Solutions (BIS) to estimate each fielder’s defensive contribution in theoretical runs above or below an average fielder at his position in that player’s league and year. Thus, a SS with a UZR of zero is exactly average as compared to a SS in the same year and in the same league. If his UZR is plus, he is above average, and if it is minus, he is below average.
Defensive Runs Saved (DRS) - Defensive runs saved is a similar idea to UZR, it is just done a little differently. There is a really good comparison at our sister site Athletics Nation written a few years ago that dives into great detail. There are some subtleties between the two (UZR ignores shifts and DRS uses them on the whole, first base positioning, etc...).
DRS is relative to that position so it only compares shortstops to shortstops, catchers to catchers, etc... If you want to move cross-position, you’ll need something like FanGraphs defensive metrics, which combines fielding runs plus a positional adjustment.
Defensive rating - This is FanGraphs core defensive metric. As noted above, it takes UZR for non-catchers and gives it a positional adjustment so you can compare players across positions (which is how you arrive at a final WAR metric). For catchers it is a bit more complicated but it uses a combination of DRS and some pitch blocking metrics to get there.
Fielding Runs Above Average (FRAA) - This is Baseball Prospectus’ defensive metric. It arrives at defensive runs a bit different than UZR/DRS (it scraps the zone-based philosophy and uses a bit more of a WOWY type). It however is not scaled to position, so catchers - the toughest position to play - make up the tops of FRAA leaderboards.
Outs Above Average - Statcast recently rolled out their version of a defensive metric. As you would expect, it uses various Statcast related data such as catch probability, distance to the wall, batted ball data, where an infielder is positioned, how far an infielder is from the base they are throwing to, how fast the runner is, and various other things. It’s certainly the most “complex” in that it takes the most minute data but that doesn’t make it the de facto best necessarily.
Unfortunately, none of the above defensive metrics are better than the other explicitly. Mainly it is because defense is really hard to judge! There is a lot of noise in an individual season. Sometimes a fielder gets a lot of really hard batted balls hit to them (a bad thing). Sometimes a hitter hardly gets any balls hit to them at all in a season, relative to others. The general rule of thumb is that is takes 2-3 years of data to figure out if a fielder is good or bad based off data. Scouting obviously helps here because you know who the really good or really bad fielders are quickly when you watch. What these metrics can do is to find the players in between, those who may be over or underrated.
Pitch Framing - Catchers sometimes “steal strikes”, meaning they turn pitches outside of the zone into strikes by positioning their glove at the catch point to “fool” umpires. This was once the talk of the town but it’s been cooled down a bit. While still an important metric, the gap between the best pitch framer and the worst has gotten closer. Catchers are paid for this metric now, so being good at this is rewarded. How long that lasts though is the question, as MLB may eventually move to automated balls/strikes, effectively eliminating all pitch framing opportunities.
The Big One
Wins Above Replacement (WAR) - No metric may have caused more discourse, bad faith, and digital vitriol than WAR. Introduced years ago, WAR takes everything about a position player (hitting, fielding, base running) and a pitcher, and turns it into one metric. One metric you can compare across time, across position, and even between hitters and pitchers. It’s dangerous to define a player by one metric, and WAR isn’t meant to be the end-all, be-all for how good/bad a player is. It does, however, track well. The best players by WAR are all basically Hall of Famers and elite players we all know. The worst players by WAR usually don’t last long in the bigs.
A replacement level player is defined essentially as “if this player was injured and I picked a freely available minor league or AAAA player (a player who is too good for AAA but not good enough for the MLB), what would be the difference.” Obviously replacement level is a bit subjective as that player doesn’t necessarily exist (sorry Jack Cust), so the idea is more theoretical. So WAR is how many wins a player is above that theoretical replacement, free available player.
I like to say don’t think of WAR in terms of certainty and specificity. A 3.2 WAR player isn’t necessarily better than a 3.0 WAR player. The 3.2 WAR player might have had a few fluky at bats or fielding attempts. The 3.2 WAR player might have had a few fly balls leave the field by only an inch or two. I’d feel comfortable at saying anything >0.5 WAR difference you can say one player was better in a season. When comparing players across time, the bands have to get a lot wider. Ted Williams wasn’t worse than Roger Hornsby despite an ~8 WAR gap between them in their careers.
I will also caution about using WAR as a matter-of-fact for players who played in the pre-data era of ~2001 when fielding metrics weren’t tracked like they are now. Players who played in the 20s and 30s whose WAR is driven a lot by defense (which we aren’t as certain of) rather than offense (which can mostly be easily calculated with some certainty) should be taken with a grain of salt and wide berths. Defensive runs have been calculated way back but the tool has been more crude. Analysts and writers basically had to dig through old box scores and look at opportunities and conversion of those opportunities into outs as defining good or bad for defense (with some general information on other player/team specific stuff). It’s nowhere nears as good as players that play in the current advanced data era.
You’ll see various ways to calculate it across the internet. FanGraphs, Baseball Prospectus, and Baseball-Reference all have their own ideas on how players should be measured. For instance for pitchers, FanGraphs uses FIP while Baseball-Reference used Runs Allowed. While the three may diverge from player-to-player in any given year, on the whole they mostly tell the same story of a player over several years.
Use WAR as a starting point for how good/bad a player is and then dive deeper into understanding how WAR gets there. Not all aspects of play are rewarded the same in free agency, even if two players have similar WAR figures. An elite fielder and below-average hitter won’t be paid the same as an elite hitter and below-average fielder, even with even WAR values.
As for defining how good a player is from a single WAR number, I again like to look at this as player bands or tiers.
Negative to 0 WAR: replacement level to you can probably DFA them and be fine
0-1 WAR: still mostly replacement level but perhaps good enough to keep on your 26-man roster
1-2 WAR: roughly average or so player
2-3 WAR: average to slightly above, but absolutely someone worth rostering
3-4 WAR: above-average player, maybe an all-star level
4-5 WAR: All-Star level, border lining on elite and MVP caliber
6-9 WAR: elite player, who is MVP caliber
9+ WAR: this player is probably Mike Trout
An important thing to remember is that of course a 26-man roster is made up of just that; 26 players. Every player takes up one spot, and those spots are finite. You’d rather have a six-win player than two three-win players, as long as the other teammate of the six-win player is above replacement level. In essence, trading a six-win player for two three-win players isn’t equal value. The six-win player may not be twice as good as those other two combined, but there is value in having that extra roster spot for additional WAR.
WAR for Relievers - While WAR for hitters and starting pitchers is mostly straightforward (as straightforward as it can get I suppose), WAR for relievers is a bit trickier. With hitters and starting pitchers, a replacement level player is fairly identifiable in theory. However, when a team’s star closer gets injured, they don’t call up a random reliever from AAA to take their place. Instead, the 8th inning pitcher becomes the closer, the 7th inning pitcher becomes the 8th inning pitcher, etc... This concept is called “bullpen chaining” and it’s fundamental to how reliever WAR is calculated. So think of replacement level for relievers not as some 0 WAR player, but the next person up in the bullpen who is getting their high leverage spots.
A truly elite reliever is worth typically 2-3 WAR per year, which as you can see above, would only be roughly an average to slightly above position player or starting pitcher. When it comes to Hall of Fame voting, this has caused a lot of turmoil but the value still makes sense. Relievers are mostly fungible (as evidenced by bullpen chaining and longevity) and they face far fewer opponents than their starting pitcher peers. In 2019, Jalen Beeks of the Rays led all relievers in total batters faced with 401. Trevor Bauer led all starters with 911, or 127% more. Beeks’ 401 batters faced would rank roughly 120th among starting pitchers that year. Having a good bullpen is important but a bullpen is much more the sum of its parts than other groups. Individual hitters and starters are often worth more than entire bullpen units via WAR.
Expected stats have been around for a bit, but with more advanced data provided by the rollout of StatCast, it’s gotten a bit different (accurate?). Rather than expected stats being driven mostly by the outcome of a batted ball or plate appearance, it is now driven by what happens prior to the end of an event. How hard the ball is hit, the launch angle, and a lot of the things we covered above can drive these expected stats.
Expected Batting Average (xAVG) - As mentioned in the paragraph right above, this is the expected batting average for a player based off how they have made contact.
Expected Slugging (xSLG) - Same for xAVG but for slugging percentage.
Expected Weighted On Base Average (xwOBA) - This one is a little trickier because it doesn’t just use what happened prior to an event because wOBA contains things like walks. So instead it’s a combination of things that actually happened (walks, hit by pitches, etc...) and then what would expected to have happened after contact (xAVG, xSLG, etc...). While wOBA itself is great, xwOBA provided a bit more flavor/predictive power because it helps separate some bad luck and noise from the signal.
wOBA - xWOBA - A lot of letters in this one but it’s a simple concept to understand: what happened vs what should have happened - theoretically at least. You can use this as a way to perhaps find over/under-performers.
If a batter (pitcher) has a wOBA (what happened) higher than xwOBA (what was supposed to happen), then it may mean that they were over(under)-performing. A batter was getting better results than they should have or a pitcher was getting worse results than they should have. Obviously, flip this for when xwOBA > wOBA (good indicator for a batter, bad indicator for a pitcher).
Expected FIP (xFIP) - Even though I’m classifying this in the “xSTATS” section, it is different than those above in that it isn’t based off StatCast data or some sort of batted ball quality information.
A common “complaint” with FIP is that it is sensitive to good/bad home run luck, given that home runs have the highest weight in the formula. What xFIP tries to do is normalize that home run circumstances by normalizing a pitchers flyball% to a league average rate, which then modifies their home runs allowed. If a pitcher has only given up a couple flyballs but they’ve all gone out of the park, then their xFIP will be lower than their FIP, with the idea that those flyballs shouldn’t turn into home runs as much going forward.
This works decently for most pitchers, but remember that there are some pitchers who just run high/low HR/FB% because of their nature.
The Very Advanced
Okay, for those of you who were really just looking to read the first couple chapters of the manual, just enough to be dangerous, you don’t have to necessarily continue on. We are going to cover some things below that you’ll probably rarely going to see discussed in amongst the broadest swath of fans.
But here is the thing...these are still important and emerging concepts/stats, so if you learn them now and get accustomed to them, you can look cool in front of your friends at parties or next to the cute person at the ballgame.
Pitch values - One thing I love about the modern era of stats is we no longer have to just live by the narrative. Years ago, if someone said “Player A has a good curveball”, we’d probably just accept it as fact and repeat it. Now, we can just go look up if Player A does have a good curveball (I regret to inform you they don’t).
Several sites offer something in this capacity: FanGraphs, Baseball Prospectus (via Brooks Baseball), and StatCast being the big ones. They all use different methods (StatCast obviously uses their data) but for the most part their values should all be close to each other. If a pitcher has a good changeup via one site, it’s likely they do at another site too.
Here is a thing to be careful on using these though: often these values are based off the final pitch. So if a pitcher throws three fastballs that go whiff, whiff, home run, on the home run will count and the prior two pitches (good pitches it seems) will be ignored.
Swing/Take Run Values - Swing/Take values are a bit hard to explain fully and briefly. Every time a pitch is thrown, the batter can choose to swing at it or not. Swinging a missing on a pitch that takes the count from 0-1 to 0-2 is worse than taking the pitch to go from 0-1 to 1-1. That’s obvious. But also perhaps not swinging at a pitch somewhere isn’t as good as swinging at a pitch somewhere.
Unfortunately, we aren’t even halfway there to describing it (maybe that is my fault) so let’s speed this up. The zone is split into four regions (heart, shadow, chase, waste) and sometimes it’s good to swing or not swing or if you swing, what happens after you swing is sometimes better depending on the zone. Swing/Take controls for that and assigns a run value to every pitch.
I like to use this to bridge the gap between the final pitch and what happened before it. Swing/Take values will take every pitch by itself rather than just that last pitch. So rather than using a pitcher/batters wOBA on fastballs, you can Swing/Take instead to account for all the fastballs.
Spin rates - While the concept of spin rates has been around for a long time, they’ve only recently really been trackable (seen Rapsodo below). There is an entire master's level coursework you could take on the physics and benefits of spin rate, how to measure it, what is best, and a bevy of other features and functions. I’m just going to link you to MLB/Statcast’s Spin Rate glossary entry, which has several other links you can read if you want to further chase the rabbit down the hole.
Active Spin - It’s probably again just best to read the Statcast entry above but active spin is the amount (expressed in percentage terms) that the spin rate of the ball contributed to the movement of the ball. The movement of a ball is dependent on two things: velocity and spin. Where movement of the ball is entirely a product of the spin rate, that is 100% active spin.
Runs Expectancy on 24 Base States (RE24) - There are 24 states in any given plate appearance: no outs/no runners, one out/no runners, no outs/one runner on first, one out/runners on first and second, etc... Add them up, and you get 24. Each of these states have historically had an expected run value. So naturally going from one base state to another, provides a plus/minus run value. If you after your plate appearance, you go from a runner on first to a runner on first and second (aka you got on base), then it increases your teams run expectancy. That would be a positive RE24 value.
This isn’t really useful for individual hitters or pitchers, as hitters/pitchers aren’t necessarily responsible for what happened prior. Hitter B shouldn’t get credit for Hitter A getting on base, even if they move him over from first to second, because then Hitter B gets some credit due to the base states.
RE24 is expressed as for the remainder of the inning. So when you see a RE24 value of say 1.45, that means at that point in the plate appearance, teams are expected to score 1.45 runs through the rest of the inning historically.
Win Probability Added (WPA) - A fairly simple one that is in just about every major sport. At any given moment, each team has a win probability. The change in win probability from one play to the next (be it a down in football, a plate appearance in baseball, a shot attempt in basketball, etc...) provides a plus/minus for win probability. Needless to say, you want a positive WPA as that means value was added - at least you want it positive for the team you are rooting for.
Leverage Index - Think a bit back to the RE24 and base states from earlier. Obviously having no runners on provides a lower run expectancy than having the bases loaded and no outs. Now take RE24 and add in the more contextual stuff like inning and score differential. So two runners on in the bottom the first in a 0-0 game is different leverage than two runners on in the bottom of the ninth up 1-0.
Leverage index is normalized a bit so it can be easily compared. An average leverage index (I think you could call this a neutral game state) is 1. Anything above 2 is considered high leverage. Anything below really 0.8ish is considered low leverage and ~60% of all leverage is under 1.
Championship Win Probability Added (cWPA) - While not a new metric, it isn’t used much but it has gained some popularity over the past few years. As you can see in the name, win probability is a big part of it, but rather than just the win probability of a single game, this measures the win probability of the entire “season” in a sense. That “season” in this case is the World Series, aka the entire reason you play.
Historically calculated/stored at Baseball Gauge, it has since been folded into the stats at Baseball-Reference.
Pitch Tunneling - With the proliferation of more advanced pitch tracking due to the introduction of Statcast, we’ve been able to quantify more things and give evidence to old concepts. Imagine I throw two pitches: a fastball and a changeup. What I want is both of these pitches to follow down the same “tunnel.” I want them to follow the same paths to the plate for as long as I can, so you have less time to decide if the pitch is a fastball or changeup. Obviously the more time a batter has to decide what pitch is currently hurling through time and space at them, the better their response time can be. If I throw both of my pitches in the same tunnel, it makes it harder for the batter. A lot goes into pitches having the same tunnel, from arm angle/release point, to movement, to spin rate...
The Projection Systems
While each projection system does something a little different, they are all mostly a function of applying a player's performance and a typical aging curve. So without going too deep into the small differences that may exist between them (often we don’t even know such things necessarily), I’ll just try to give a little background on them.
And as always, no system is perfectly accurate. Some have been historically more accurate than others, but as any system dealing with humans, they are fallible (as are the humans).
ZiPS - Created by Dan Szymborski many moons ago, ZiPS stands for sZymborski Projection System. A cool thing about ZiPS is that Dan provides daily updates (though they aren’t run at the same full course as the annual projections) and longer term projections over at FanGraphs,
PECOTA - The system every fan loves to hate, PECOTA was created by Nate Silver now of FiveThirtyEight. PECOTA (which stands for Player Empirical Comparison and Optimization Test Algorithm - a backronym referring to Bill Pecota) is available at Baseball Prospectus.
Clay Davenport - Clay’s projections aren’t available at any of the Big Three stat sites (FanGraphs, Baseball-Reference, Baseball Prospectus), they are housed at his own webpage. They are incredibly detailed and give several different iterations of projection levels.
Marcel - I love the idea of the Marcel system. Tom Tango (now of MLB Advanced and Statcast guru) wanted the most simple system but one that also provided what he called the “minimum competency” of any projection system. It uses very basic inputs and weighting.
It can be found now at Baseball-Reference.
Steamer - While not trying to be offensive to its creators (Jared Cross, Dash Davidson and Peter Rosenbloom), Steamer is a fairly simple system overall, with some added blend of pitch/batted ball data built in. And even in its simplicity, it is one of the more accurate projection systems.
FanGraphs Depth Charts- While FG DC isn’t it’s own system, it does give a sort of a wisdom of the crowd. DC combines ZiPS and Steamer into one to arrive at the projection. The key thing though is that the playing time projections are manually edited, given the wealth of knowledge at the site from its contributors. It gives an accurate estimation of playing time that is updated continually. Something mostly no other system does.
ATC - Created by Ariel Cohen, ATC is mostly fantasy-focused. It is similar to Depth Charts due to manual playing time modifications and a blend of projections. However, unlike DC where it is a pure blend of ZiPS and Steamer, ATC actually weights each stat by historical accuracy for a projection system in projection that stat. So any projection system that has historically projected home runs well, gets a higher weight for the home run projection piece.
THE BAT and THE BAT X - Also fantasy-focused, THE BAT uses numerous inputs for its projections for both annual and daily. However, the daily projections has even more minute data like the umpire, air density, etc... for that game. THE BAT X then folds in Statcast data to the projection.
The Way to Value Things
$/WAR - The central thesis of Moneyball (aka: sabermetrics) is that you should be buying wins. What $/WAR tries to do is put a numeric value on both what a win has cost and what you are paying for a win. The lower the $/WAR value, the more efficient your spending has been. A team that wins 100 games and spends only $100M to do so, was more efficient than a team that wins 105 games but spends $200M to do so. However, does it matter? In the end, not really. There really should be no ceiling for what you would pay to win five more games than the next team because that is ultimately what matters. However, there is an argument that the team that paid $100M for 100 wins has more upward mobility in potential spending when considering the de facto salary cap that the luxury tax acts as.
We can also use $/WAR as a framework for contracts and what a team would/should be willing to pay a player. Historically, the cost of a win has ranged recently ~$7 million. So if a player is projected for 2-wins in a season, the breakeven contract for that player for that one year is ~$14M. Teams build models based off this to know what to pay for a player going forward. Obviously the further out you go, the less certainty you have, and anything greater than 2-3 years forward you are weighting aging curves very heavily. This isn’t to say all teams base every dollar they pay for a player off this. There are still negotiations, nuance, and uncertainty in models. Still, this framework can give a guide to how to value a player.
Future Value - Somewhat similar to $/WAR, future value is used to assign a singular number to a player. However, as its name might imply, it is reserved for prospects. The seminal piece on FV was done over at FanGraphs a few years ago, which I urge you to read here. The idea being to assign a number, based off the 20-80 scouting scale, to a prospect for like-to-like purposes.
Rapsodo - At one point, fans and teams who wanted detailed pitch information were subject to whether or not they had access to advanced pitching systems that costs hundreds of thousands of dollars to install. Then along came a small device, no bigger than bucket, that allowed anyone (who could afford it) to have detailed pitch tracking. They’ve since expanded to include also hitting metrics and into the sport of golf as well. Beyond just the physical device, what Rapsodo has done the most is expand the discussion and advancement of detailed stats and metrics. Maybe it hasn’t changed the way things are analyzed, but it has at least brought the discussion into the minds of far more people than before it.
Statcast - While MLB has had ball tracking information installed at stadiums for about a decade prior (known as Pitch F/X), it wasn’t until they rolled out Statcast that things took a quantum leap. While there was a lot of information that Pitch F/X gave us, Statcast took us from implied information to measured information. All of the new concepts in baseball are likely brought about by something Statcast measures, be it exit velocity, launch angle, spin rate, etc... Statcast is available now at every MLB park and some spring training stadiums too. I believe currently there are some minor league parks with Statcast data, it isn’t public like the way it is with MLB stadiums.
Hawkeye - In mid-2020, MLB did away with their old Statcast technology (provided by Trackman) and replaced it with Hawkeye technology. This is the same tech that is used in a sport like tennis, where it is most famous for being the tech that tells if a ball was in/out of bounds. With the technology change came new advancements in what can be tracked as well. Hawkeye moves beyond just ball tracking, and tracks player movements and kinetics as well. While the kinetics piece is still mostly untapped, player tracking has allowed for Statcast to provide things like defensive metrics, sprint speed, and route efficiency.
The Useful Sites
FanGraphs - While not the original sabermetrics site, it is now the leader (in my opinion). FanGraphs was started by David Appleman in 2005 and has since gone on to being at the forefront of pushing the analytical discourse. Countless contributors to the site over the years have gone on to join roles at teams, the media, and even a few working directly for MLB and the MLBPA.
Baseball Prospectus - Baseball Prospectus (or BP as it is more colloquially known) has been around since ~1996. Most notable probably for being the owner of the PECOTA projections, it also is arguably the best site on the internet for the discussion of baseball in general. While they house a ton of data, I think they contribute the most to the social and ethical side of baseball more than any other site.
Statcast - MLB’s signature state page, their leaderboards and gamefeeds show data in real time.
Baseball-Reference - The OG of baseball stats - created by Sean Forman in 2000 - started out as just a college kid (working on his Ph.D) wanting to make stats more available and easier to find. Along the way they started creating their own metrics and rolled out the invaluable Play Index tool. Now it houses data across multiple sports across the globe (American football, hockey, basketball, association football).
Clay Davenport - Clay’s site doesn’t get as much love as it should. Clay runs and creates his own projection systems which are just as detailed as any other system and even gives several different passes/iterations as outputs the users can see. From long term projections, percentiles, and closest comparable players, Davenport translations deserve a spot in the echelon of popular systems.
Pitcher List - Created initially in 2014 by Nick Pollack as just a way to show how filthy pitchers are by posting GIFs, Pitcher List has grown into an entire media/data site that is still a church to the religion of the gods of pitching.
Beyond the Box Score - Under the umbrella of Royals Review media parent in Vox, BsBS has been around for...well, I’m not sure but for a long time. It runs and looks similar to us here at Royals Review but obviously covers all of the MLB landscape and analytically tilted.
The Baseball Cube - TBC was created by Gary Cohen back in 2003. Today it catalogues more detailed history of a players stats than probably any other site. TBC has for some players their: high school, college, minors, and majors stats. It also carries biographical information for players, the history of the MLB draft, and MLB prospect rankings. It’s an incredibly deep source of information and one of the few “mom and pop” baseball sites still out there.